Effect of Friction Stir Processing on Hardness and Microstructure in La-Based Bulk Metallic Glass Composites

Abstract

Friction stir processing (FSP) was conducted on a 2.5 mm thick La₆₆Al₁₄Cu₂₀ bulk metallic glass composites (BMGCs) plate. Effects of FSP on the BMGCs have been investigated. Microstructural observations indicated that the crystalline particles were elongated and parallel to the rotating direction of the pin after FSP. Micro-hardness measurement showed that hardness of the stirred zone drops down approximately 50 Hv compared with the as-cast sample. The mechanical properties of the amorphous matrix before and after FSP have been investigated using nanoindentation test. Results shown that creep displacement-time curves of the BMGCs before and after FSP can be accurately fitted by the generalized Kelvin model which was usually used in polymeric materials. The creep compliance and creep retardation spectrum shown that the as-cast specimen is in a more relaxed state.

Fig. 3 SEM micrograph of the La₆₆Al₁₄Cu₂₀ specimen (a) before and (b) after FSP. (c) Same as for micrograph (b) but at lower magnification. (d) Typical SEM micrograph around the indents after indentation measurements for the processed sample and the corresponding high-resolution image (inset).

1. Introduction

There are a larger number of papers researching on severe plastic deformation of crystalline materials. For example, FSP may refine the microstructure and contribute to improve mechanical properties of crystalline materials.^1–3) In deed, Arora et al.⁴⁾ also found that the average size of the dendrites of Ti-based BMGCs was reduced by almost one fifth by FSP. But no other researches on severe plastic deformation of BMGCs were found. Therefore, extensive and in-depth research on FSP of BMGCs is needed. After all, it is the BMGCs who can basically meet the requirements of engineering applications, due to the fact that some BMGCs can not only exhibit the special mechanical, chemical and physical properties of single-phase bulk metallic glasses (BMGs), but also can overcome the brittleness problem of single-phase BMGs at room temperature.⁵⁾

The material used in this study is one of the best known glass formers in the La-based system, La₆₆Al₁₄Cu₂₀, with very low glass transition temperature of 391 K and a large supercooled liquid region of 65 K.⁶⁾ A relatively wide operating window and a low working temperature are necessary for the FSP without auxiliary heating. FSP was conducted on a 2.5 mm thick La₆₆Al₁₄Cu₂₀ BMGCs plate. A smooth nugget zone without obvious defects was achieved under the tool rotation speed of 1000 rpm, welding speed of 15 mm/min, and the plunge depth of 0.3 mm.

The purpose of this paper is to study the effect of FSP on the microstructure evolution and mechanical properties of the nugget zone, including the changes of the grains and the amorphous matrix. Plenty of researches have indicated that nanoindentation is an ideal technique to investigate the microscopic plastic deformation of BMGs due to its high accuracy.^7–9) Therefore, the nanoindentation creep tests were conducted to investigate the mechanical property changes of the amorphous matrix in the BMGCs during FSP. The microstructure evolution of the crystalline phase of the nugget zone before and after FSP was also observed.

2. Experimental Procedure

Plates of La₆₆Al₁₄Cu₂₀ BMGCs with a dimension of 60 × 20 × 2.5 mm³ were prepared by water-cooled copper mold suction casting method under an Ar atmosphere. The friction stir processing was performed on a computer numerical control vertical milling machine (FSW-3LM-015). The rotating tool used in this study was made of tool steel and was a cylindrical shoulder with 12 mm in the diameter. The frustum cone-like pin part was 2 mm and 3 mm in the diameter of upper and lower sides respectively. The length of the pin is 2 mm.

The microstructures of the specimens were examined by QUANTA FEG 450 scanning electron microscope (SEM) and X-ray diffraction (XRD, Rigaku D/max-RB) with Cu-Kα radiation at a scanning rate of 2°/min and a detecting step of 0.02°. The differential scanning calorimetric (DSC, STA449C) was used to determine glass transition temperature (T_g = 396 K) and crystallization temperature (T_x = 446 K) at a heating rate of 20 K/min under the protection of high purity Ar gas. The supercooled liquid region, ΔT_x = T_x − T_g, of 50 K. Micro Vickers hardness test was carried out at an applied load of 500 gf and dwell time of 10 s to compare the mechanical properties of the samples before and after FSP.

The nanoindentation creep tests were conducted at room temperature on a Nanotest600 nanoindenter utilizing a diamond Berkovich indenter. We will specify the nanoindentation creep behaviors of the amorphous matrix in the BMGCs before and after FSP. The indentation creep tests were loaded to 10 mN with different loading rates of 0.075 mN/s, 0.2 mN/s and 1 mN/s, and then held for 10 s to evaluate their creep behaviors. Thermal drift rate was maintained below 0.05 nm/s during each test.

3. Results

Figure 1 shows XRD patterns of the as-cast La₆₆Al₁₄Cu₂₀ specimen and the stir zone of the processed sample. XRD pattern of the as-cast sample shows tiny crystalline peaks of body centered cubic phase LaAl₄ and trigonal system phase CuLaO₂ phases embedded on a broad halo peak, characteristic of the amorphous composites. The absorption of atomic oxygen may be happened in the storing process of La materials, which can be identified by the color change of La material. The peaks of LaAl₄ and CuLaO₂ phases in the XRD patterns of the processed sample seem much sharper than in the as-cast sample, indicating the occurrence of crystallization during the FSP.

Fig. 1

XRD patterns of the as-cast specimen and stir zone of processed sample.

Figure 2 shows the micro-hardness distribution across the cross section perpendicular to the welding direction, at the centre in the thickness direction. The zero point represents the centre of the stir zone. The average micro-hardness of the as-cast sample is 246 Hv. However, the micro-hardness of the stirred zone drops down approximately 50 Hv, and the value is low to 186 Hv in the centre of the stir zone. It has been reported that, the micro-hardness of the stir zone may have no marked difference compared with the base material for the fully amorphous metallic glass when the glass phase had no significant change,¹⁰⁾ or increases when a crystalline layer was formed.¹¹⁾ Therefore, there must be some microstructure changes in the bulk metallic glass composites.

Fig. 2

Micro Vickers harness profile of the FSP region in the as-cast and processed samples.

4. Discussions

4.1 Microstructure evolution of the crystalline phase during the FSP

Figure 3 shows the SEM micrograph of the La₆₆Al₁₄Cu₂₀ BMGCs before and after FSP. Crystalline particles of LaAl₄ phase and CuLaO₂ phase with a size of 1–5 µm were found to be homogeneously dispersed in the amorphous matrix of the as-cast sample (Fig. 3(a)). But we can only see CuLaO₂ phase in Fig. 3. The LaAl₄ phase nearly can’t be seen in the SEM micrograph because of the similar darkness to the amorphous matrix. In contrast, although there is no significant change in the size of the crystalline particles between the samples before and after FSP, the crystalline particles are elongated and parallel to the rotating direction of the pin after FSP (Fig. 3(b)). Furthermore, a lot of cavities are presented in the amorphous matrix. Besides, the cavities have an increasing trend toward the centre direction of the stir zone (Fig. 3(c)), which may be responsible for the decrease of the micro-hardness of the processed sample. It is well established that cavitation occurs in most polycrystalline materials during superplastic deformation, and cavitation has an effect on mechanical properties of the materials after deformation.^12,13) The reason and effect of cavitation on BMGs and BMGCs deserve further investigation. Figure 3(d) shows the typical SEM micrograph of the surface obtained after indentation. Though a large number of particles exist in the composites, shear bands parallel to the imprint edge are revealed and no obvious signs of circular shear bands and piling-up can be observed, which indicate the brittleness of the composites. Additionally, it can be clearly seen from the internal high-resolution image that shear bands initiate from the particles, by pass the cavities and eventually propagate into the amorphous matrix, which indicate that the first yielding point determined by the particles. Therefore, the overall micro-hardness of the processed samples may be related to the cavities, but indeed depends on the particles and the amorphous matrix.

Fig. 3

SEM micrograph of the La₆₆Al₁₄Cu₂₀ specimen (a) before and (b) after FSP. (c) Same as for micrograph (b) but at lower magnification. (d) Typical SEM micrograph around the indents after indentation measurements for the processed sample and the corresponding high-resolution image (inset).

4.2 Changes of the amorphous matrix during FSP

To further understand the microstructure evolution dependence on the free volume in the amorphous matrix during FSP, a nanoindentation test was conducted. And all the indenters were located on the amorphous matrix in every nanoindentation test for all the specimens before and after FSP. Figure 4 shows the nanoindentation load-displacement (P-h) curves of the as-cast and the processed samples. There are obvious indentation creep phenomena in all the curves during peak-load holding segment.

Fig. 4

Load-displacement curves of (a) as-cast and (b) processed samples.

The creep displacement versus time (h-t) during the peak-load holding period with various loading rates for the as-cast and processed specimen are plotted in Fig. 5. The starting points of creep displacement are zeroing for comparison. All curves, denoting the creep process, can be segmented into two stages approximately: the transient creep and the steady-state creep. In the first (transient creep) stage, the creep displacement increases rapidly with the holding time, and the creep strain rate decreases with time. In the second (steady-state creep) stage, the increase rate of creep displacement becomes tardiness with the holding time and even closes to a constant. It is clearer that the increase in penetration depth during the load-holding segment in the as-cast sample is much more pronounced than in the processed sample. This implies that severe structural relaxation occurred due to the heat input in the FSP process, making the content of free volume decrease drastically and enhanced the creep strength and creep resistance.

Fig. 5

Creep displacement versus time during the peak-load holding period with various loading rates for (a) the as-cast and (b) processed specimen.

Furthermore, the maximum creep displacements increase with the increase of the loading rate from 0.075 mN/s to 1 mN/s both for the as-cast and processed specimen. Similar trend has been reported on other BMGs.^14–18) The above phenomena are closely related to the viscoelastic and viscous flow of BMGs. With a higher loading rate, corresponding to a shorter loading time, the time-dependent viscoelastic and viscous flow processes are suppressed during the loading stage, and fully developed during the following peak-load holding stage, so that the obvious creep deformation appears; while at a lower loading rate, there is sufficient time for the BMGCs to viscoelastic and viscous flow during the loading stage, so the creep deformation is not prominent during the peak-load holding stage. Thus, the dependence of nanoindentation creep displacement on loading rate is attributed to the delayed plastic deformation to some extent. Furthermore, according to Spaepan’s free volume theory,¹⁹⁾ a large number of excess free volume can be produced during the deformation process. Since the positive correlation between the production rate of free volume and the strain rate, a large number of excess free volume can be produced when it was loaded at a high rate. Together with the original free volume without enough time to be annihilated, a large number of free volume at the high loading rate create conditions for the rearrangement of atoms, which leads to the decrease of creep resistance. So the creep deformation at a high loading rate becomes more prominent during the peak-load holding stage, and vice versa.

Considering that the maximum creep displacement during the peak-load holding stage can minimize the viscoelastic deformation in the loading stage, the loading rate of 1 mN/s was chosen to further analyze the creep behavior of the BMGCs before and after FSP. Figure 6 shows the h-t curves for the BMGCs before and after FSP. The generalized Kelvin model was used to fit the creep displacement curves.²⁰⁾ In Kelvin model, the viscoelastic behavior of BMGs can be described with a battery of dashpots and linear springs and the creep displacement h in the peak-load holding stage can be expressed as   

\begin{equation} h(t) = h_{e} + \sum_{t = 1}^{n}h_{i}(1 - e^{-t/\tau_{i}}) + \frac{t}{\mu_{0}} \end{equation} (1)

where h_e is the indentation depth at the first spring; h_i and τ_i represent the indentation depth and the retardation time for the ith Kelvin element; μ₀ is a constant related to the viscosity coefficient of the last dashpot. Two exponential terms in eq. (1) is enough to get a accurate fit for the BMGs in this study. The fitting parameters are listed in Table 1. The retardation times in the creep behavior of polymers under tensile tests, τ₁ and τ₂, represent α and β transitions, which probably are associated with the main-chain movement and molecular characteristic of the polymers, respectively.²¹⁾ If the same idea is adopted into the indentation creep tests of BMGCs, the retardation times in Table 1, τ₁ is much smaller than τ₂ for both the two specimens, may related to the chemical short-range order structural relaxation and the topological short-range order structural relaxation, respectively.

Fig. 6

Typical experimental fitted displacement-time curves during the peak-load holding period with the loading rate of 1 mN/s for (a) the as-cast and (b) processed specimen.

Table 1 Young’s modulus (E), hardness (H) and the fitting parameters in eq. (1) for the two specimens.

Creep compliance J(t) is an important mechanical physical quantity that reflects the properties of materials. Based on the generalized Kelvin model, the creep compliance can be induced from the fitted parameters, which can be approximately expressed as (employing two exponential terms):   

\begin{equation} J(t) = \frac{A_{0}}{P_{0}h_{\textit{in}}}h(t) \end{equation} (2)

where A₀ and P₀ are the contact area and the applied load corresponding to the virtual length, respectively; The virtual length h_in, approximately equals the depth at the end of the loading.

Using eq. (2) and fitted parameters in Table 1, the relationship between holding time and creep compliance of the two specimens are shown in Fig. 7. The creep compliance curves for the two specimens nearly have the same shape. The creep compliance curves are initially constant and then dramatic increases with holding time, which is caused by the structural relaxation during the peak-load holding stage.²²⁾ Similar phenomena have been observed in Ti,²²⁾ La,²³⁾ Ce²³⁾ and Cu^24,25) based BMGs. Generally, the constant creep compliance for Ti,²²⁾ La,²³⁾ Ce²³⁾ and Cu^24,25) based BMGs is approximately equal to10⁻¹⁰ Pa⁻¹, obviously consistent with the results of this experiment. Furthermore, comparing the two figures, it can be seen that the creep compliance of the processed specimen is smaller than that of the as-cast specimen. Since the increasing compliance is often referred to as softening dispersion,²³⁾ the as-cast specimen with large compliance value indicates its more relaxed state, which also reflects that severe structural relaxation had occurred during the FSP. The increase of the creep compliance means the decrease of Young’s modulus and hardness. Therefore, the processed specimen should have a larger Young’s modulus than that of as-cast specimen. Table 1 listed the tested hardness and Young’s modulus of the two specimens, the calculation results are in good agreement with Table 1.

Fig. 7

Creep compliance of (a) as-cast specimen and (b) processed specimen under a maximum load of 10 mN at a loading rate of 1 mN/s.

Based on the eq. (2), the creep retardation spectrum was deduced, which can be approximately expressed as (employing two exponential terms):   

\begin{align} L(\tau) &= \frac{dJ(t)}{d\ln t} - \frac{d^{2}J(t)}{d(\ln t)^{2}}\bigg|_{t = 2\tau} = \left[\left(1 + \frac{t}{\tau_{1}}\right)\frac{h_{1}}{\tau_{1}}e^{-t/\tau_{i}} \right.\\ &\quad \left. + \left(1 + \frac{t}{\tau_{2}}\right)\frac{h_{2}}{\tau_{2}}e^{-t/\tau_{2}}\right]\frac{A_{0}}{P_{0}h_{\textit{in}}}t\bigg|_{t = 2\tau} \end{align} (3)

The retardation spectra of the as-cast specimen and the processed specimen are shown in Fig. 8. It is discovered that every retardation spectrum is made up of two peaks with well-defined relaxation times. And the second peak is much broader than the first peak for both the two specimens. In order to exclude the influence of the exponential terms, we increase the exponential terms from two to three. Interestingly, the number of the peaks in the retardation spectrum still is two, which indicates that there are two different kinds of relaxation processes during the indentation creep process. The same phenomenon has also been observed in Ti,²²⁾ La,²³⁾ Ce,^20,23) Cu^24,25) and Mg²⁶⁾ based BMGs. Furthermore, the relatively sharper retardation peaks are observed in the as-cast sample. This indicates that the as-cast sample is in a more relaxed state, which is in consistent with the results of creep compliance.

Fig. 8

Retardation spectra of as-cast and processed specimen under a maximum load of 10 mN at a loading rate of 1 mN/s.

Therefore, it is reasonable to suppose that the overall free volume in the amorphous matrix of the BMGCs decreased during the FSP, due to the severe structural relaxation caused by the heat input, even though a large number of free volumes may be produced by the deformation process in the FSP.

5. Conclusion

The microstructure evolution and mechanical properties of La₆₆Al₁₄Cu₂₀ BMGCs before and after FSP have been investigated. The main findings of this study are:

1. (1)    The volume fraction of LaAl₄ and CuLaO₂ phases in the processed La₆₆Al₁₄Cu₂₀ BMGCs increased compared with the as-cast samples.
2. (2)    The crystalline particles are elongated and cavities are presented in the amorphous matrix after FSP,
3. (3)    Although superplastic deformation of BMGs in the supercooled liquid region always generates free volumes, FSP of La₆₆Al₁₄Cu₂₀ BMGCs brings structural relaxation and cavities to the amorphous matrix.
4. (4)    The micro-hardness of the processed BMGCs drops down approximately 50 Hv compared with the as-cast samples, due to the increased volume fraction of particles and the cavitation of the matrix.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (NSFC) [Grant No. 51661015 and Grant No. 51661016]; and the Natural Science Foundation of Gansu Province [Grant No. 1606RJZA050].

REFERENCES

• 1)   A.P.  Reynolds: Scr. Mater. 58 (2008) 338–342.
• 2)   C.I.  Chang,  C.J.  Lee and  J.C.  Huang: Scr. Mater. 51 (2004) 509–514.
• 3)   Y.J.  Kwon,  I.  Shigematsu and  N.  Saito: Scr. Mater. 49 (2003) 785–789.
• 4)   H.S.  Arora,  S.  Mridha,  H.S.  Grewal,  H.  Singh,  D.C.  Hofmann and  S.  Mukherjee: Sci. Technol. Adv. Mater. 15 (2014) 035011.
• 5)   X.Y.  Zhang,  Z.Z.  Yuan,  X.L.  Feng,  L.Z.  Cui and  D.X.  Li: Mater. Sci. Eng. A 620 (2015) 352–358.
• 6)   Q.C.  Zhang,  S.J.  Pang,  Y.  Li and  T.  Zhang: Sci. China Phys. Mech. 54 (2011) 1608–1611.
• 7)   C.A.  Schuh: Mater. Today 9 (2006) 32–40.
• 8)   B.C.  Wei,  L.C.  Zhang,  T.H.  Zhang,  D.M.  Xing,  J.  Das and  J.  Ecker: J. Mater. Res. 22 (2007) 258–263.
• 9)   T.  Burgess,  K.J.  Laws and  M.  Ferry: Acta Mater. 56 (2008) 4829–4835.
• 10)   Y.S.  Ji,  H.  Fujii,  M.  Maeda,  K.  Nakata,  H.  Kimura,  A.  Inoue and  K.  Nogi: J. Phys. Conf. Ser. 165 (2009) 012015.
• 11)   J.  Kobata,  Y.  Takigawa,  S.W.  Chung,  H.  Tsuda,  H.  Kimura and  K.  Higashi: Mater. Trans. 48 (2007) 2409–2413.
• 12)   S.  Tekeli and  M.  Erdogan: Ceram. Int. 28 (2002) 785–789.
• 13)   B.P.  Kashyap: Surf. Interface Anal. 31 (2001) 547–559.
• 14)   J.  Sort,  J.  Fornell,  W.  Li and  S.S.  Baró: J. Mater. Res. 24 (2009) 918–925.
• 15)   Y.H.  Chen,  J.C.  Huang,  L.  Wang and  T.G.  Nieh: Intermetallics 41 (2013) 58–62.
• 16)   F.  Xu,  Z.L.  Long,  X.H.  Deng and  P.  Zhang: Trans. Nonferrous Met. Soc. China 23 (2013) 1646–1651.
• 17)   G.K.  Liao,  Z.L.  Long,  M.S.Z.  Zhao,  L.  Peng,  W.  Chai,  X.J.  Xu,  Z.H.  Ping,  Y.  Zhang and  C.P.  Yan: Mater. Res. Express. 4 (2017) 115202.
• 18)   B.G.  Yoo,  J.H.  Oh,  Y.J.  Kim,  K.W.  Park,  J.C.  Lee and  J.  Jang: Intermetallics 18 (2010) 1898–1901.
• 19)   B.G.  Yoo,  K.W.  Park,  J.C.  Lee,  U.  Ramamurty and  J.  Jang: J. Mater. Res. 24 (2009) 1405–1416.
• 20)   B.C.  Wei,  T.H.  Zhang and  W.H.  Li: Mater. Trans. 46 (2005) 2959–2962.
• 21)   S.  Yang,  Y.W.  Zhang and  K.Y.  Zeng: J. Appl. Phys. 95 (2004) 3655–3666.
• 22)   X.  Wang,  P.  Gong,  L.  Deng,  J.S.  Jin,  S.B.  Wang and  P.  Zhou: J. Non-Cryst. Solids 470 (2017) 27–37.
• 23)   W.H.  Li,  K.  Shin,  C.G.  Lee,  B.C.  Wei,  T.H.  Zhang and  Y.Z.  He: Mater. Sci. Eng. A 478 (2008) 371–375.
• 24)   J.L.  Wu,  Y.  Pan and  J.H.  Pi: Physica B 421 (2013) 57–62.
• 25)   J.  Wu,  Y.  Pan and  J.  Pi: J. Mater. Eng. Perform. 23 (2014) 486–492.
• 26)   Y.  Sun,  Z.Q.  Li,  J.  Liu,  M.Q.  Cong and  J.Y.  Qin: J. Rare Earth. 29 (2011) 253–258.